RECOVERY OF 29 SECOND OSCILLATIONS IN HUBBLE-SPACE-TELESCOPE ECLIPSE OBSERVATIONS OF THE CATACLYSMIC VARIABLE UX URSAE-MAJORIS

Low-amplitude (similar or equal to 0.5%) 29 s oscillations have been d etected in Hubble Space Telescope Faint Object Spectrograph eclipse ob servations of the nova-like cataclysmic variable UX UMa. These are the same dwarf nova-type oscillations that were originally discovered in this system by Warner & Nather in 1972. The 29 s oscillations are seen in one pair of eclipse sequences obtained with the FOS/PRISM in 1994 November but not in a similar pair obtained with the FOS/G160L grating in August of the same year. The oscillations in the PRISM data are si nusoidal to within the small observational errors and undergo an appro ximately -360 degrees phase shift during eclipses (i.e., one cycle is lost). The amplitudes are highest at preeclipse orbital phases and exh ibit a rather gradual eclipse whose shape is roughly similar to, altho ugh perhaps slightly narrower than, UX UMa's overall light curve in th e PRISM bandpass (2000-8000 Angstrom). Spectra of the oscillations hav e been constructed from pre-, mid, and posteclipse data segments of th e November observations. The spectra obtained from the out-of-eclipse segments are extremely blue, and only lower limits can be placed on th e temperature of the source that dominates the modulated flux at these orbital phases. Lower limits derived from blackbody (stellar atmosphe re) model fits to these data are greater than or equal to 95,000 K (gr eater than or equal to 85,000 K); the corresponding upper limits on th e projected area of this source are all less than 2% of the white dwar f (WD) surface area. By contrast, oscillation spectra derived from mid eclipse data segments are much redder. Fits to these spectra yield tem perature estimates in the range 20,000 K less than or similar to T les s than or similar to 30,000 K for both blackbody and stellar atmospher e models and corresponding projected areas of a few percent of the WD surface area. These estimates are subject to revision if the modulated emission is optically thin. We suggest that the ultimate source of th e oscillations is a hot, compact region near disk center, but that a s ignificant fraction of the observed, modulated flux is due to reproces sing of the light emitted by this source in the accretion disk atmosph ere. The compact source is occulted at orbital phases near mideclipse, leaving only part of the more extended reprocessing region(s) to prod uce the weak oscillations that persist even at conjunction. The highly sinusoidal oscillation pulse shape does not permit the identification of the compact component in this model with emission produced by a ro tating disturbance in the inner disk or in a classical, equatorial bou ndary layer. Instead, this component could arise in a bright spot on t he surface of the WD, possibly associated with a magnetic pole. Howeve r, a standard intermediate polar model can also be ruled out since UX UMa's oscillation period has been seen to change on timescales much sh orter than the minimum timescale required to spin up the WD by accreti on torques. A model invoking magnetically controlled accretion onto di fferentially rotating WD surface layers may be viable, but needs more theoretical work.